Browsing by Author "Ladd, Helen F"

The purpose of this project was to investigate the extent to which Communities In Schools (CIS) of Durham is using the appropriate eligibility criteria for admission into its graduation coach program. The study provides information that could help the organization better target at-risk students for coaching services by answering the following research question: Is CIS of Durham’s eligibility methodology appropriate for selecting students into its graduation coach program?

II. Description of CIS of Durham

Created in 1992, CIS of Durham is a member of Communities In Schools, one of the largest dropout prevention networks in the United States. The organization uses the graduation coach program to help at-risk students stay in school and graduate. The program includes college tours, mentoring, skills development, and internships (level II services) for a select number of students. The program also provides school-wide services, including basic needs resources and health screenings (level I services). All services are voluntary.

CIS of Durham targets students for level II services that it has identified as being at risk of dropping out. The organization selects students using attendance, academics, and misbehavior measures. The organization divides these predictors into risk levels (see Table 1). Any student who falls in the Low, Medium, or High risk levels in two out of the three predictors qualifies for the program.

III. Study Design

I used data from the North Carolina Education Research Data Center (NCERDC) for descriptive and logistic analyses of the graduation coach program’s eligibility criteria. I identified 10th grade students from the Durham Public Schools in the 2009-2010 academic year and determined if they had dropped out by the end of their senior year. Descriptive analysis: I divided the student sample into three groups: all 10th graders, students eligible for the program, and dropouts. For each group, I provided summary statistics on the percent of students in every risk level for the three CIS of Durham indicators. I also described each group’s ethnic and gender composition. Logistic analysis: I used logistic regression (logit) to determine if the CIS of Durham indicators are good predictors of dropping out. The logit estimates the probability that an event will occur. I controlled for gender, ethnicity, and whether or not a student received free/reduced price lunch.

I divided each CIS of Durham indicator into categories that correspond to the risk levels the organization uses to admit students into the program. This division allowed me to determine if the odds of dropping out are greater at higher risk levels.

IV. Findings & Recommendations

A. Descriptive statistics

I identified 1,845 10th graders from five traditional Durham schools in 2009-2010. A total of 110 students dropped out at least once by the end of their senior year. Under the CIS of Durham eligibility criteria, 276 students (15%) qualified for the program. Ethnicity: Eligible students and dropouts tended to be minorities. While 56% of 10th graders were black, 73.2% of eligible students belonged to this group. 62% of dropouts were black and 21% were Hispanic students. White students comprised only 7% of dropouts even though they made up 26% of the sample.

Gender: About 11% more males qualified for the program than females. Also, 6% more males dropped out than females.

Absences: 77% of all 10th graders had fewer than 15 absences (Outside risk) in 2010-2011. 73.2% of eligible students missed fewer than 18 days (Outside-Low risk), while 46% of dropouts were absent more than 25 days (Not Included risk).

Suspensions: 75.5% of 10th graders had no suspensions (Outside risk). By comparison, nearly 76.5% of eligible students had 1-4 suspensions (Low-High risk). For dropouts, as the risk levels increased, the percent of students in each level decreased, except for the Not Included level. The data showed a substantial jump in the number of dropouts with 5 or more suspensions.

GPA: 62% of 10th graders had a C average grade or better (Outside risk). 76% of eligible students had a GPA between 1.01 and 2.0 (Low-Medium risk). 44.5% of dropouts had a GPA below 1.0, indicating that these students tend to perform worse academically than their peers.

B. Logit results

The logistic analysis indicated that CIS of Durham could enhance the program’s admission criteria. I recommend eliminating the two-out-of-three rule by dropping suspensions since the data demonstrated that absences and suspensions had a redundant effect on dropping out. Gender, ethnicity, and free/reduced price lunch showed no significant effect.

The data showed that the GPA ranges of the risk levels are correctly denominated for this predictor. The odds of dropping out increase as GPA decreases. I propose no changes to this indicator. On the other hand, the ranges of the risk levels for absences could be improved. In particular, the Medium and High risk categories did not have a significant effect on the odds of dropping out compared to some of the other levels. I recommend changing the ranges as shown in Table 2. The suggested new ranges are better denominated since higher absences risk levels are associated with statistically significant higher odds of dropping out.

V. Further Considerations

The results of this project cannot be generalized beyond the design of the study and its assumptions. CIS of Durham should treat the study’s recommendations as a general rule of thumb. Graduation coaches should make every effort to assess the individual needs of students.

In this dissertation, I explore the impact of several public policies on civic participation. Using a unique combination of school administrative and public–use voter files and methods for causal inference, I evaluate the impact of three new, as of yet unexplored, policies: one informational, one institutional, and one skill–based. Chapter 2 examines the causal effect of No Child Left Behind’s performance-based accountability school failure signals on turnout in school board elections and on individuals’ use of exit. I find that failure signals mobilize citizens both at the ballot box and by encouraging them to vote with their feet. However, these increases in voice and exit come primarily from citizens who already active—thus exacerbating inequalities in both forms of participation. Chapter 3 examines the causal effect of preregistration—an electoral reform that allows young citizens to enroll in the electoral system before turning 18, while also providing them with various in-school supports. Using data from the Current Population Survey and Florida Voter Files and multiple methods for causal inference, I (with my coauthor listed below) show that preregistration mobilizes and does so for a diverse set of citizens. Finally, Chapter 4 examines the impact of psychosocial or so called non-cognitive skills on voter turnout. Using information from the Fast Track intervention, I show that early– childhood investments in psychosocial skills have large, long-run spillovers on civic participation. These gains are widely distributed, being especially large for those least likely to participate. These chapters provide clear insights that reach across disciplinary boundaries and speak to current policy debates. In placing specific attention not only on whether these programs mobilize, but also on who they mobilize, I provide scholars and practitioners with new ways of thinking about how to address stubbornly low and unequal rates of citizen engagement.

This dissertation examines the interplay between school district assignment and reassignment policies and the elementary public school parents select for their children. The sample in all chapters includes the third and fourth grade students in a subset of growing North Carolina school districts from 2003/04 to 2010/11. The data are derived from historical, longitudinal secondary data sources containing student, school, and district records. All chapters employ quantitative longitudinal data analysis methods. Chapter 1 identifies the groups of students who do not comply with their school assignments. Chapter 2 identifies the groups of students who are reassigned to different schools, and to schools of varying quality, when school districts enact reassignment plans. Chapter 3 identifies the groups of students who do not comply with school reassignments. Together, the chapters demonstrate the interplay between residential decisions, school choices, and the resulting educational opportunities of observably different students. Consistent with existing bodies of literature, the findings demonstrate unexplored processes through which advantaged families maintain the most desirable educational opportunities for their children. Policy implications of these findings are also discussed.

This dissertation consists of three studies exploring the effects of natural disasters in North Carolina on the longer term outcomes of children. The first study looks at the effect of prenatal natural disaster exposure on maternal health behaviors and birth outcomes for twenty cohorts of children born in North Carolina. Combining North Carolina administrative and survey data on births with disaster declarations from the Federal Emergency Management Agency (FEMA) allows me to identify children who were exposed to disasters in each trimester of prenatal development. Using a county fixed effect strategy, I compare these children to other children born in the same county who were not exposed to disasters while in utero. Results indicate that prenatal natural disaster exposure, especially exposure to hurricanes, has a significant effect on some maternal health behaviors, but this study provides only limited support for the theory that natural disaster exposure negatively affects birth outcomes, as measured by birth weight and gestational age.

The second study looks at the impact of exposure to natural disasters during pregnancy on the educational outcomes of North Carolina children at third grade. A broad literature relates negative birth outcomes to poor educational performance, and a number of recent studies examine the effect of prenatal exposure to natural disasters on birth outcomes. This study takes the next step by considering how prenatal exposure affects later outcomes. The children identified in the first study as exposed to disasters prenatally are compared to other children born in the same county who were not exposed to disasters while in utero. Results suggest that children exposed to hurricanes prenatally have lower scores on third grade standardized tests in math and reading. Those exposed to flooding or tornadoes also have somewhat lower math scores. Additionally, results suggest that these negative effects are more concentrated among children in disadvantaged subgroups, especially children born to Black mothers.

The third study addresses the question of whether the disruption caused by a natural disaster has an impact on student academic outcomes in the school year during which the natural disaster occurs. The effects of disasters on school performance are important because natural disasters often constitute a major community disruption with widespread impacts on the lives of children. The educational data in this study comes from administrative records for all school districts in North Carolina. Results suggest that hurricanes have a negative overall impact on reading test scores, with the effect concentrated among middle schools. However, winter storms have a positive effect on both math and reading scores in middle school. This difference in effect and additional analysis of mechanisms suggests that mobility is more important than missed days of schools in mediating negative effects of hurricanes on school performance.